|Publication number||US7545815 B2|
|Application number||US 10/967,591|
|Publication date||Jun 9, 2009|
|Filing date||Oct 18, 2004|
|Priority date||Oct 18, 2004|
|Also published as||CA2523351A1, CA2523351C, DE602005008331D1, EP1648125A1, EP1648125B1, US20060083226|
|Publication number||10967591, 967591, US 7545815 B2, US 7545815B2, US-B2-7545815, US7545815 B2, US7545815B2|
|Original Assignee||At&T Intellectual Property Ii, L.P.|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (24), Referenced by (1), Classifications (15), Legal Events (2)|
|External Links: USPTO, USPTO Assignment, Espacenet|
The present invention relates to a queueing technique and, more particularly, to a queueing technique for equitably handling multiple sources, with multiple priorities, entering a single queue.
There are many applications, particularly within the realm of data transmission, where multiple streams of information need to be collected and routed through various points. For example, routers are used with the Internet to move data traffic from one point to another across the network. As is well-known in the art, each data packet may include an identified “class of service” (COS) that defines the priority with which that packet is to be moved with respect to other packets in the network. Obviously, as millions of these packets are moved every minute, “traffic jams” at various router locations are likely to occur.
Complicating matters is the every-increasing use of virtual private networks across the IP network. A virtual private network (VPN) is a private data network that makes use of the public telecommunications infrastructure, maintaining privacy through the use of a tunneling protocol and security procedures. A virtual private network can be contrasted with a system of owned or leased liens that can only be used by one company. The idea of the VPN is to give the company/owner the same capabilities at a much lower cost by using the shared public infrastructure rather than a private one. When multiple VPNs use a common router (which is often the case), it is possible for the high priority traffic from one VPN to interfere with the high priority traffic associated with the other VPNs.
Thus, a remaining problem in the art is that there is currently no guarantee of “fairness” among high priority traffic flows from different VPNs sharing a common router.
The need remaining in the prior art is addressed by the present invention, which relates to a queueing technique and, more particularly, to a queueing technique for equitably handling multiple sources, with multiple priorities, entering a single queue.
In accordance with the present invention, a technique referred to as “round robin” weighted fair queueing (WFQ) has been developed that not only addresses the queueing issues previously corrected by using a WFQ approach, but further implements the ability to pre-determine the VPN order for use with the WFQ to ensure that high priority traffic from each VPN is treated in similar fashion. Multiple queues are utilized, each associated with a different VPN, with an additional queue including tokens identifying the particular COS associated with the next packet to be added to the data stream. A queue selector, configured to identify the next queue to insert a packet into the data stream, is used in association with the COS queue to provide the desired “round robin” function to the queueing technique. As a result, the inventive round robin weighted fair queueing (RR-WFQ) preserves the service level for a particular class as a whole, but alters the queueing of packets from each VPN subscribing to that class.
In a further arrangement of the present invention, the amount of traffic generated by each VPN can be included with the RR-WFQ to further even out the availability of the data channel to each VPN. That is, if one VPN generates twice as much traffic as two other VPNs (all being of the same COS), the queue selector will be configured so that the first VPN will be queued up twice in each “round robin” cycle. This advanced arrangement of the present invention is defined as “weighted round robin, weighted fair queueing” (WRR-WFQ).
It is to be understood that although the details of the queueing methodology of the present invention are described below in terms of queueing multiple classes of packet traffic from multiple VPNs sharing a single server, the technique itself is applicable to virtually any situation where a “single element” needs to equitably deal with multiple sources of information/data, with multiple priority levels associated with the data at each source.
Indeed, the general methodology of the present invention will become apparent during the course of the following discussion and by reference to the accompanying drawings.
Referring now to the drawings,
As discussed above, the ability to equitably support multiple VPNs (usually with competing traffic of the same “high” priority) through a single router (or any other gating device) is difficult in today's data network architecture. One approach that is currently used to address this situation is “weighted fair queueing”, where each different class of service (COS) is assigned a different weight. Table I, as shown below, is exemplary of one weighting scheme:
Class of Service
WFQ Weight (w)
It is presumed for the purposes of the present discussion that each VPN sharing a single router generates traffic at the COS2 level. When using the WFQ technique, the scheduled departure time, Fj, for each arriving packet, P, is determined by using as inputs both the arrival time, t0, of the packet and the length, L, of the packet. If there are currently no packets of that class in the queue, the scheduled departure time is calculated by the following:
F j =t 0 +L/w i,
where wi is the WFQ weight for that particular COS. If there are previous packets of that class already in the queue, the departure time of the newly-arrived packet is defined by:
F j =F j−1 +L/w i,
where Fj−1 is defined as the departure time for the previous packet of that same COS in the queue.
The application of this prior art WFQ technique to scheduling traffic from a pair of VPNs is illustrated in
Class of Service (COS)
As shown in
The problem of sharing a singe router with multiple VPNs (or in the more general case, for sharing a single “sink” for information with multiple “sources”, each “source” having multiple priorities), is addressed by the round-robin weighted fair queueing technique of the present invention. A simplified implementation of the RR-WFQ technique is illustrated in
In accordance with the present invention, a separate priority queue, denoted QP, is included in the arrangement, where queue QP holds tokens associated with the COS levels of each traffic packet, with the highest level COS tokens disposed in the beginning of the queue. In this case, a set of four COS2 tokens (for packets P1, P2, P4 and P5) are disposed in front of the COS3 token associated with packet P3. A queue selector QS is used in conjunction priority queue QP to determine the identity of the particular queue to be selected for offering a packet. As will be discussed below, the queue selector QS may simply cycle through each queue, one at a time, until each queue has had a turn. Alternatively, different schemes and selection algorithms may be applied.
Referring back to
As a result of implementing the round robin strategy of the present invention with the set of WFQ-implemented VPN queues, packet P5 does not have to wait for P1, P2 and P4 to all be transmitted before it is selected (since otherwise the length of P5 would push it behind P4, as shown in
The round robin situation described above (taking one packet from each queue in turn) works well if the amount of traffic generated within each queue is ideally one-to-one. However, in some (if not most) of the cases, one VPN will generate significantly more or less traffic than the other VPNs sharing the same router. In this case, the simple round robin technique would disadvantage those VPNs generating more traffic by treating each VPN the same. Thus, an extension of the inventive technique can be used to “weight” each queue feeding into router, where this technique is referred to as “weighted round robin, weighted fair queueing”, or WRR-WFQ.
As shown in
An important aspect of the RR-WFQ technique of the present invention is that it preserves the service level for a particular COS as a whole (through the priority queue), simply by altering the service sequence within that class. Moreover, the scheme is simple enough to be implemented in application-specific integrated circuits (ASICs), as is required by the desire to develop a low-cost high-speed solution. Indeed, both of the RR-WFQ and WRR-WFQ techniques of the present invention are considered to be significant improvements over the prior art WFQ so as to allow for multiple VPNs (in general, multiple users of a single entity) to fairly share a class of service level at a single connection point into the network. Both of these inventive queueing protocols are, like WFQ, simple enough to implement that they may easily be incorporated into today's routers.
It is to be understood, as mentioned above, that the present invention goes well beyond the implementation of servicing multiple VPNs through a single router. Indeed, the present invention is deemed to be applicable to any situation where multiple sources of information are to be funneled into a single information sink, where various items of information may have different priorities. Thus, the inventive RR WFQ techniques of the present invention are considered to be limited in scope only by the claims appended hereto.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US5959993 *||Sep 13, 1996||Sep 28, 1999||Lsi Logic Corporation||Scheduler design for ATM switches, and its implementation in a distributed shared memory architecture|
|US5999534 *||Nov 24, 1997||Dec 7, 1999||Daewoo Electronics Co., Ltd.||Method and apparatus for scheduling cells for use in a static priority scheduler|
|US6182177 *||Jun 13, 1997||Jan 30, 2001||Intel Corporation||Method and apparatus for maintaining one or more queues of elements such as commands using one or more token queues|
|US6198723 *||Apr 14, 1998||Mar 6, 2001||Paxonet Communications, Inc.||Asynchronous transfer mode traffic shapers|
|US6201793 *||Mar 16, 1998||Mar 13, 2001||Lucent Technologies||Packet delay estimation in high speed packet switches|
|US6317416 *||Oct 11, 1996||Nov 13, 2001||Alcatel Canada Inc.||Fair queue servicing using dynamic weights (DWFQ)|
|US6359861 *||Aug 10, 1998||Mar 19, 2002||Massachusetts Institute Of Technology||Method for scheduling transmissions in a buffered switch|
|US6438134 *||Aug 24, 1998||Aug 20, 2002||Alcatel Canada Inc.||Two-component bandwidth scheduler having application in multi-class digital communications systems|
|US6452933 *||Nov 18, 1997||Sep 17, 2002||Lucent Technologies Inc.||Fair queuing system with adaptive bandwidth redistribution|
|US6538989 *||Sep 9, 1998||Mar 25, 2003||British Telecommunications Public Limited Company||Packet network|
|US6594704||Dec 15, 1999||Jul 15, 2003||Quarry Technologies||Method of managing and using multiple virtual private networks in a router with a single routing table|
|US6671258||Feb 1, 2000||Dec 30, 2003||Alcatel Canada Inc.||Dynamic buffering system having integrated random early detection|
|US6744767||Dec 30, 1999||Jun 1, 2004||At&T Corp.||Method and apparatus for provisioning and monitoring internet protocol quality of service|
|US6744776||Oct 18, 2000||Jun 1, 2004||Advanced Micro Devices, Inc.||Servicing priority traffic in multiport network switch|
|US6862265 *||Apr 13, 2000||Mar 1, 2005||Advanced Micro Devices, Inc.||Weighted fair queuing approximation in a network switch using weighted round robin and token bucket filter|
|US7054267 *||Sep 10, 1999||May 30, 2006||Lucent Technologies Inc.||Method and apparatus for scheduling traffic to meet quality of service requirements in a communication network|
|US7110411 *||Mar 25, 2002||Sep 19, 2006||Erlang Technology, Inc.||Method and apparatus for WFQ scheduling using a plurality of scheduling queues to provide fairness, high scalability, and low computation complexity|
|US7142513 *||May 23, 2002||Nov 28, 2006||Yea-Li Sun||Method and multi-queue packet scheduling system for managing network packet traffic with minimum performance guarantees and maximum service rate control|
|US7190674 *||Oct 25, 2002||Mar 13, 2007||Fujitsu Limited||Apparatus for controlling packet output|
|US7257083 *||Jun 25, 2003||Aug 14, 2007||Motorola, Inc.||Method and apparatus for policy-based dynamic preemptive scheduling of data transmissions|
|US7342929 *||Apr 26, 2002||Mar 11, 2008||Cisco Technology, Inc.||Weighted fair queuing-based methods and apparatus for protecting against overload conditions on nodes of a distributed network|
|US20030103514 *||Mar 5, 2002||Jun 5, 2003||Hong-Soon Nam||Apparatus and method for packet scheduling using credit based round robin|
|US20030112755 *||Dec 17, 2001||Jun 19, 2003||Worldcom, Inc.||Virtual private network (VPN)-aware customer premises equipment (CPE) edge router|
|US20040093492||Nov 13, 2002||May 13, 2004||Olivier Daude||Virtual private network management with certificates|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US20130283097 *||Apr 23, 2012||Oct 24, 2013||Yahoo! Inc.||Dynamic network task distribution|
|U.S. Classification||370/395.42, 370/358, 370/230.1, 370/386|
|Cooperative Classification||H04L47/6225, H04L12/5693, H04L47/621, H04L47/245, H04L47/522|
|European Classification||H04L12/56K, H04L47/62D1, H04L47/62B, H04L47/52B, H04L47/24E|
|Oct 18, 2004||AS||Assignment|
Owner name: AT&T CORP., NEW YORK
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:CAHN, ROBERT;REEL/FRAME:015911/0295
Effective date: 20041013
|Oct 4, 2012||FPAY||Fee payment|
Year of fee payment: 4